home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: sci.math.symbolic
- Path: sparky!uunet!charon.amdahl.com!pacbell.com!sgiblab!sdd.hp.com!saimiri.primate.wisc.edu!zaphod.mps.ohio-state.edu!magnus.acs.ohio-state.edu!usenet.ins.cwru.edu!agate!stanford.edu!rock!concert!samba!steve
- From: steve@physics.unc.edu (Steve Christensen)
- Subject: Re: Serious Programming (was: Re: MAPLE resources reccomendation)
- Message-ID: <1992Nov18.043011.28800@samba.oit.unc.edu>
- Sender: usenet@samba.oit.unc.edu
- Nntp-Posting-Host: augustus.physics.unc.edu
- Organization: MathSolutions, Inc. - Chapel Hill, NC.
- References: <Bxtxzy.CJD@news.cso.uiuc.edu> <BxuxC6.Dxz@news.cso.uiuc.edu> <1eb8j4INNld2@agate.berkeley.edu>
- Date: Wed, 18 Nov 1992 04:30:11 GMT
- Lines: 67
-
- With all due respect to the recent contributors to this
- line of postings [all of whom I consider to be extremely
- interesting, helpful, and knowledgable researchers] ...
-
- I normally try to stay out of such "religious" discussions, but this time
- I cannot resist. I am noting comments like
-
- "Pattern matching is nice, but..."
-
- "It's ok for fiddling around, but..."
-
- with regard to Mathematica's abilities. We have written a 250 function
- commercial piece of software that contains - in a very modular form -
- over 25,000 lines of Mathematica code and comments. Before we did this
- - starting in the late 80's - we tried virtually every system available
- with great care. [I even learned enough LISP to help with the porting
- of the KCL version of Maxima to the Sparc compilers.]
- Given the complexity of the functions we needed,
- relying almost totally on pattern matching, only Mathematica had the
- resources we needed. Not being experts at the deep theory of
- programming language syntax, we plowed ahead and wrote our software. I
- can honestly say that for me programming in Mathematica was very
- straightforward and even fun. So far, we have been able to devise
- every kind of structure and function that we needed. When a bug or
- feature of Mathematica caused some problem early on, we were always
- able to find a work around that did not require any C programming. We
- have been extremely happy with the ease with which we are able to
- create new functions.
-
- We have discussed with some of Mathematica's competitors about porting
- our code to their software. The first question we have to ask is - can
- your software do all the kinds of pattern matching that Mathematica
- can? So far, the answer has been - "No, but we are planning to do
- that." If this has changed, I would love to hear from one of my
- friends developing the other systems.
-
- We have no desire to write in C or LISP since this would require us to
- have compilers for maybe 15 different kinds of computer architecture or
- operating system, not to mention the hardware. With Mathematica we do
- all our development on our Sun workstations and the code runs on all
- machines that run Mathematica.
-
- It is true that Mathematica does not work by loading a lot of libraries
- (though the newer "thin" versions do some of that), but I have found
- that systems that load a lot of libraries can get very boring to use.
- [I recently have been studying Axiom, which I find interesting, but I
- got rather annoyed when it took about 5 minutes to load a bunch of
- libraries before an integral could be done. When I did the same thing
- in Mathematica, it took a few seconds. Perhaps I am still too much of
- an amateur with Axiom.]
-
- I would be very happy to hear details on just how wonderful some other
- system is at doing pattern matching. I consider powerful pattern
- matching to be by far the most important feature of a computer
- math system. With it, I can teach the system to do just about anything
- I want. Other systems may have some programming syntax that is clearer
- to a programming theorist, but I could care less. Some systems might
- be able to do some integral that Mathematica can't, but again I could
- care less - I can teach Mathematica to do integrals if I like.
-
- Perhaps I am missing some of the finer points of the the discussions
- in this news group, but I think that some of it should be in
- some group like comp.software.religion --- probably including this
- posting. Oh, well... Amen.
-
- Steve Christensen
- MathSolutions, Inc.
-